FastAPI Playground

A project where I make FastAPI do some cool stuff such as generic filtering, async database use and more.

It really takes the effort out of developing APIs.

Interesting bits

These are some interesting takeaways to maybe copy and paste into your own projects.

File	What does it do
playground/providers/database.py	This module provides sync sessions for the database
playground/providers/database_async.py	This module provides async sessions to talk to the database
playground/providers/settings.py	Automatically read env vars and expose them to us with defaults
playground/routers/auditing.py	Automatically do auditing-related things with requests. Any request will have extra logging applied.
playground/routers/auth_passthrough.py	Create a HTTP client with a users credentials so we make requests on a users behalf.
playground/routers/health.py	Implement some extremely simple ready and live checks.
playground/routers/paginator.py	Paginate in-memory data in a re-usable manner.
playground/routers/time_range.py	Filter and request data from a database with filtering implemented in a re-usable manner.
tests/conftest.py	Creates pytest fixtures for the database, async database and a testclient. It can then be used in the tests without effort.
tests/mimesis.py	A small test case to test fake data with mimesis.
tests/mimesis.py	Run tests repeatedly and with parameterized instances.
tests/routers/test_*.py	Test API endpoints, backed by an in-memory SQLite database.
tests/test_schematesis.py	Automatically test the entire API from the spec without writing any cases!

Why is FastAPI so neat

It's a library, not a framework, making it easy to integrate it into any environment
Hot reload that takes less than a second to restart
Pydantic is used for data parsing, making it easy to describe datamodels
OpenAPI spec is generated automatically from Pydantic models and FastAPI routes
Open-ended dependency system that makes dependencies a breeze to manage
Fully type annotated

What are some other cool tools in here

Hypothesis

Hypothesis is a data generator for test cases. It can generate random strings, numbers and even Pydantic models! This makes is very useful to automatically write unit tests without setting up a gazillion test cases.

Schemathesis

Built on top of Hypothesis, schemathesis allows us to automatically test an entire API. We can point it at an OpenAPI spec and it will immediately try every endpoint. If errors occur or the endpoint responds outside of its documentation, schemathesis will throw it as an error.

Pytest extensions

pytest-fast-first > Runs the fastest tests first and the slowest tests last
pytest-clarity > Improve error messages by providing diffs
pytest-snapshot > Write and validate tests with snapshots. Useful when there are a lot of validations to do
pytest-repeat > Run a test multiple times. Useful when dealing with randomness
respx > Mock HTTP servers to test API calls

Black

Black in a code-formatter that does not let you configure it. It's for the best.

How to run it?

First make sure you have the following installed:

python (3.10 or higher)
poetry (https://python-poetry.org/). This tool installs dependencies and arranges virtual-envs

Now we can run the following

poetry install
poetry run uvicorn playground.main:app --reload --host 0.0.0.0 --port 8000

You can now open http://127.0.0.1:8000 in your browser. Uvicorn will automatically reload if you make any changes.

What about docker?

Of course it runs in docker as well. Simply run docker-compose up -d to start it.

What about code quality tools

poetry run pflake8 # Linting
poetry run mypy playground # Validate type annotations
poetry run pytest # Run unit and api tests

Requinard/fastapi-playground